14591

2009, 113, 1459114594

Published on Web 07/27/2009

Simultaneous Oxidation and Reduction of Arsenic by Zero-Valent Iron Nanoparticles:

Understanding the Significance of the Core-Shell Structure

Mauricio A. V. Ramos, Weile Yan, Xiao-qin Li,, Bruce E. Koel, and Wei-xian Zhang*,
Center for AdVanced Materials and Nanotechnology, Department of Chemistry and Department of CiVil and
EnVironmental Engineering, Lehigh UniVersity, Bethlehem, PennsylVania 18015
ReceiVed: June 2, 2009; ReVised Manuscript ReceiVed: July 20, 2009

Increasing evidence suggests that nanoscale zerovalent iron (nZVI) is effective for the removal of arsenic
from contaminated water, but the immobilization mechanism is unclear. In particular, the existence of As(0)
on the nanoparticle surface has been proposed but not substantiated in prior studies. By using high-resolution
X-ray photoelectron spectroscopy (HR-XPS), we report clear evidence of As(0) species on nZVI surfaces
after reactions with As(III) or As(V) species in solutions. These results prove that reduction to elemental
arsenic by nZVI is an important mechanism for arsenic immobilization. Furthermore, reactions of nZVI with
As(III) generated As(0), As(III), and As(V) on the nanoparticle surfaces, indicating both reduction and oxidation
of As(III) take place with nZVI treatment. The dual redox functions exhibited by nZVI are enabled by its
core-shell structure containing a metallic core with a highly reducing characteristic and a thin amorphous
iron (oxy)hydroxide layer promoting As(III) coordination and oxidation. Results demonstrated here shed light
on the underlying mechanisms of arsenic reactions with nZVI and suggest nZVI as a potential multifaceted
agent for arsenic remediation.

Introduction nZVIs strong reducing capabilities toward chlorinated hydro-

Arsenic is a well-known carcinogen, and arsenic contaminated
groundwater is prevalent around the world, particularly in South
Asia. In the United States, groundwater supplies about one-
third of the countrys drinking water. However, there are many
locations in the southwestern states where groundwater contains
arsenic concentrations in excess of the World Health Organiza-
tion (WHO) guidelines and the U.S. Environmental Protection
Agency (USEPA) drinking water standard of 10 g/L.1-3 On a
worldwide scale, it has been estimated that over 137 million
people in more than 70 countries are affected by arsenic
poisoning of drinking water.4
The principal forms of arsenic in natural waters are arsenate
[As(V)] and arsenite [As(III)]. Arsenate exists as oxyanions
(H2AsO4- or HAsO42-) in a pH range of 2-12, while arsenite
remains as neutral undissociated species (H3AsO3) below a pH
of 9.2.5,6 As(III) oxidation to As(V) by dissolved oxygen alone
is kinetically slow.4,7 Traditional remediation technologies rely
largely on adsorption for removing arsenic in water using
materials such as iron oxides, activated alumina, granular
activated carbon, and silica with various degrees of success.9,10
Recent studies suggest that nanoscale zerovalent iron (nZVI),
a highly reactive engineered nanomaterial extensively evaluated
for groundwater remediation and hazardous waste treatment, is
effective for arsenic removal in water, but the reaction mech-
anisms of arsenic on nZVI are not well understood. Because of
* Corresponding author. Phone: (610) 758-5318. Fax: (610) 758-6405.
E-mail: wez3@lehigh.edu.
Department of Chemistry.
Department of Civil and Environmental Engineering.
Present address: College of Environmental Science and Engineering,
South China University of Technology, Guangzhou 510006, P.R. China.
10.1021/jp9051837 CCC: $40.75
carbons and numerous inorganic contaminants, it is expected
that both arsenic(V) and arsenic(III) are amenable to reduction
by nZVI. Prior publications indicate, however, the transforma-
tion of As(V) to As(III) is kinetically slow, and no clear evidence
for the formation of elemental arsenic [As(0)] has been published
so far.5,11,12 On the other hand, several recent studies suggest
that, instead of reduction, oxidation of arsenic(III) is the
prevalent reaction under oxic conditions through peroxide or
radicals generated by Fenton reactions.8,13 The majority of these
studies rely on solution phase arsenic analysis, and there is a
lack of studies investigating arsenic speciation on nZVI surfaces
that can provide more direct evidence of the reactions between
nZVI and arsenic.
The objective of this work was to examine surface reactions
of arsenate and arsenite with nZVI by using high-resolution
X-ray photoelectron spectroscopy (HR-XPS), which can provide
valuable information about the valence states and abundance
of various arsenic species on nanoparticle surfaces. The results
are interpreted in light of the fundamental structure of nZVI,
and the implications of these findings on arsenic treatment and
remediation are discussed.
Materials and Experimental Methods
The iron nanoparticles (nZVI) used in this study were
prepared from sodium borohydride reduction of ferric iron as
reported previously.14,15 The nanoparticles have a mean diameter
of 60 nm and a BET-measured surface area of 30 m2/g.14
Reagent grade sodium arsenate or arsenite was used to make
As(III) or As(V) stock solutions. To eliminate dissolved O2 as
a potential oxidant in these solutions and to simulate prevalent
anoxic conditions encountered in groundwater environments,
2009 American Chemical Society
14592 J. Phys. Chem. C, Vol. 113, No. 33, 2009
Figure 1. As3d HR-XPS spectra of (a) fresh nZVI, (b) nZVI reacted
with 100 mg/L As(V), and (c) nZVI reacted with 100 mg/L As(III).
The nZVI loading in parts b and c was 5 g/L. The proportions of the
respective species as percentages of the total As detected are annotated
on the spectra. The intensity scale varies for each curve.
the arsenic solutions were purged with high-purity N2 for 30
min prior to adding nZVI. HR-XPS analysis was performed on
reacted nZVI residues using a Scienta ESCA 300 X-ray
photoelectron spectrometer. Detailed experimental procedures
and details concerning XPS analysis including spectral decon-
volution are available in the Supporting Information.
Results and Discussion
Figure 1 displays XPS spectra of the As3d region of nZVI
before (a) and after reactions with As(V) (b) and As(III) (c)
species, respectively. The XPS peak positions were assigned
by comparison to As3d5/2 binding energies reported in the
literature.16 No arsenic was detected on pristine nZVI, but As
peaks were clearly detected on reacted nZVI. The XPS spectra
reveal the presence of multiple arsenic valence states, viz.,
As(V), As(III), and As(0) on the nZVI surfaces. Curve fitting
of the As peaks involved six curves, since the As3d peak for
each chemical state has two unresolved components due to
spin-orbit splitting corresponding to the As3d3/2 (at higher BE)
and As3d5/2 (at lower BE) peaks separated by 0.7 eV (held
constant herein). The relative abundance of each chemical
species can be quantified from the respective peak areas. In
addition, information on the total As surface composition
obtained from XPS is summarized in Table S1 of the Supporting
Information, and these data are used below to infer the relative
extents of different reactions that occurred on nZVI surfaces.
Reactions with As(V). When nZVI (5 g/L) was added to
100 mg/L arsenate [As(V)] solution, significant amounts of
As(III) and As(0) were detected on the nZVI surface in 24 h
(Figure 1b), establishing As(V) reduction to As(III) and As(0).
The amounts indicated in Figure 1 are the proportions of the
respective arsenic species as percentages of the total arsenic
detected. It should be noted that each of these spectra was not
affected by increasing the X-ray irradiation time by a factor of
6 under the instrument conditions used in this study; i.e., XPS
analysis had no observable effect on the arsenic valence state
detected. We have also analyzed nZVI reacted with arsenic for
Letters
Figure 2. (a) Bright field TEM micrograph of nZVI nanoparticles
showing a core-shell structure comprised of a dense metallic center
and a thin, continuous oxide skin. The inset shows the electron
diffraction pattern of the metallic core, which suggests a polycrystalline
bcc structure. (b) HR-TEM micrograph of a single nZVI nanoparticle.
The lack of long-range periodic lattice fringes in the oxide layer suggests
that it is amorphous in character.
up to 15 days, and the results indicate no significant change in
arsenic distributions compared to samples reacted for 24 h under
the same conditions. While earlier studies reported the reduction
of arsenate to arsenite species by iron fillings or nanoscale ZVI
with considerably slower kinetics (e.g., >30 days before As(III)
was detected),6,11 the formation of As(0) has only been suggested
but not substantiated, and the predominant mechanism for As(V)
uptake is generally thought to be surface sorption and
precipitation.11,17 The results presented in Figure 1 clearly show
that nZVI is capable of reducing As(V) with reasonably fast
kinetics and reduction to elemental arsenic is also a feasible
mechanistic step. As will be discussed later herein, the variances
in environmental conditions, in particular arsenic concentration
and solution pH, and the properties of iron materials play crucial
roles in determining the final arsenic speciation.
Reactions with As(III). When nZVI was added to an arsenite
[As(III)] solution (Figure 1c), 51% of the surface-bound arsenic
remained as As(III). Meanwhile, 14 and 35% of the total arsenic
emerged as As(V) and As(0), respectively, indicating that both
As(III) oxidation and reduction had occurred to significant
extents. While As(III) oxidation by ZVI was noticed in prior
studies,5,22 the concomitant observation of oxidation and reduc-
tion of arsenic(III) has not been reported before. This behavior
implies that the nanoparticles used here have a more complex
character than being a monofunctional reductant as widely
acknowledged.
To understand the properties of nZVI, advanced transmission
electron microscopy (TEM) techniques were used to analyze
the material constituents present in individual nZVI particles.
Bright field TEM imaging showed a composite layered structure
comprised of a dense metallic center enclosed by a thin layer
of iron oxide material (Figure 2a). Phase-contrast TEM imaging
did not reveal periodic lattice fringes in the oxide layer, which
suggests a disordered oxide structure (Figure 2b). The oxide
outer shell appears to vary from 10 to 20 nm in thickness and
has a stoichiometric composition of FeOOH, as determined
previously from Fe2p and O1s XPS spectra.18 The oxide layer
is thought to form instantaneously upon nZVI synthesis to
passivate the highly reactive Fe0 core, and it may be practically
viewed as an inherent part of the nanoparticle.19 The extremely
thin dimension and disordered nature of the oxide layer support
earlier notions that the oxide phase is able to allow electron
passage via tunneling effects or defect sites,15,19 thereby
conserving the reducing characteristics of Fe0 and accounting
for As(III) reduction. On the other hand, iron oxides have been
extensively studied to possess strong adsorptive ability for both
Letters
Figure 3. As3d HR-XPS spectra from nZVI particles exposed to
different initial concentrations of As(III). The nZVI loading was 5
g/L in all cases. The inset of part a is a vertically expanded view of
spectrum a.
As(III) and As(V), and certain iron oxides formed from Fe0
corrosion are able to cause As(III) oxidation to As(V).5,20 In
our studies, the nanoparticles are comprised of an Fe0 core and
iron oxide at the interface between Fe0 and aqueous solution.
The two phases are present in a nanoscale core-shell config-
uration that allows each component to exert its distinctive
reactivity without significant kinetic constraints. The composite
particles therefore possess dual properties of the two compo-
nents, and this leads to unconventional chemical behavior such
as the concurrent oxidation, reduction, and sorption of As(III)
illustrated in Figure 1c.
Effect of As(III) Concentrations. The effect of arsenite
concentration on the product speciation has not been examined
before. Figure 3 presents results following reaction of nZVI with
arsenite at different initial As(III) concentrations. The initial
As(III) concentration in the solution was in the range 50-1000
mg/L. Clearly, the relative amount of As(0) formed is the
greatest at the lowest initial concentration of As(III) (50 mg/
L), with approximately one-half of the total arsenic present as
As(0). The proportion decreased to 35% when the initial As
(III) concentration was increased to 100 mg/L, and no As(0)
was detected when the As(III) concentration was increased to
500 and 1000 mg/L.
The solution Eh and pH conditions corresponding to different
As(III) concentrations are presented in Table S2 of the Sup-
porting Information. The equilibrium pH changed from circum-
neutral (pH 8.5) to alkaline (pH11) as the As(III) concentration
increased from 50 to 1000 mg/L. The equilibrium pH was
largely influenced by the starting pH, which showed an
increasing trend as the As(III) initial concentration increased
due to arsenite hydrolysis and the release of hydroxyl ions
(AsO2- + H2O f H3AsO3/H2AsO3- + OH-). The alkaline
condition associated with high Ar(III) concentrations promotes
ferric ion precipitation. This point is corroborated by the O1s
XPS spectra of the reacted samples (Figure S2, Supporting
Information), which show a pronounced increase in the OH-
peak size with increased initial arsenic loading. Iron oxide
precipitates are also evident in scanning electron microscopy
J. Phys. Chem. C, Vol. 113, No. 33, 2009 14593
(SEM) images (Figure S4, Supporting Information), where the
sample after reaction with 1000 mg/L As(III) exhibited extensive
platelet formation resembling typical solid precipitates from iron
corrosion (Figure S4c, Supporting Information). In contrast, the
oxide growth was less predominant in the 100 mg/L sample
(Figure S4b, Supporting Information). A further piece of
evidence in support of oxide buildup comes from solution Eh.
The Eh potentials registered a sharp increase with As(III)
concentration at 500 and 1000 mg/L (Table S2, Supporting
Information), suggesting the prevalence of iron oxidation
products observed in HR-XPS and SEM at those high concen-
trations forms a barrier layer on top of metallic iron and
diminishes the reducing capability of Fe0. Taken together, As(III)
concentration exerts a pronounced effect on As(III) reduction
by controlling solution chemistry and influencing the nZVI
surface conditions. Formation of As(0) was more evident at
lower concentrations, and it was absent at above 100 mg/L due
to substantial iron oxidation and precipitate formation. Although
the concentration range evaluated here is higher than arsenic
occurrence in natural waters (typical range 1-5000 g/L),4 the
trend revealed in Figure 3 strongly implies that reduction of
arsenite at trace levels such as in natural waters is likely to occur
in the presence of nZVI. As(0) not being observed in previous
studies may be attributed to instrumental limitations and the
specific reaction conditions used. In those studies, relatively high
As(III) concentrations were used, since conventional XPS
analysis requires a significant amount of arsenic deposited on
the particle surfaces to be detected. Under those conditions,
As(0) formation is not favored for the reasons just mentioned
above, and surface saturation with adsorbed arsenic species may
attenuate the already weak As(0) signal. In the present HR-
XPS study, we used a Scienta ESCA 300 instrument, which
has an intense rotating anode X-ray source and enhanced signal
detection utilizing a 300 mm radius hemispherical analyzer and
position-sensitive detector, thereby greatly lowering the arsenic
detection limit and improving the energy resolution. This enables
us to study nZVI reactions with arsenic at lower initial
concentrations, which are conditions more relevant to the natural
systems.
Compared to reduction of As(III) to As(0), oxidation of
As(III) to As(V) is less sensitive to the initial As(III) concentra-
tion (Figure 3). In all cases, As(V) exists as a minor species on
the nanoparticle compared to the other two valence states.
As(III) oxidation by nZVI has been reported in previous studies,
although the exact mechanism remains uncertain. It is purported
that more than one oxidant exists, including peroxide or radical
species generated from Fe0 corrosion, hydroxide species, and
iron oxides.5,20-22 Peroxide- or oxygen-containing radical gen-
eration requires dissolved O2 in an acidic medium21,22 and is
not likely to be involved considering our reaction conditions.
As(III) oxidation in the aqueous phase by hydroxyl ions is
plausible,20 but the resultant As(V) species in the solution phase
is unlikely to reattach to the nZVI surface due to surface charge
repulsion between As(V) and the oxide layer at neutral to
alkaline pH.23 It is more likely that As(III) was oxidized by
iron oxide, which proceeded via formation of iron oxide-As(III)
surface complexes.5 The lack of a distinct trend for As(III)
oxidation within the concentration range studied is probably a
result of several interacting effects, since the affinity of As(III)
for iron oxides, the amount of iron oxides available as adsorptive
sites, and the oxidation potential of iron oxides are intricately
affected by pH.
The evidence for the simultaneous occurrence of oxidation,
reduction, and adsorption of As(III) on the nZVI surface
14594 J. Phys. Chem. C, Vol. 113, No. 33, 2009
suggests that the two components of nZVI are able to bring
about different reactions in parallel, which is an interesting
property that has not been noted before. A time-dependent study
of arsenic speciation on the nZVI surface (Figure S3, Supporting
Information) provides a clue to the relative kinetics of the
respective reactions. As shown in Figure S3 in the Supporting
Information, As(0) accumulation is relatively slow in compari-
son to As(III) adsorption and oxidation. It may also be implied
from the figure that the products formed initially, e.g., As(V),
may be reduced back to As(III) and then to As(0); hence, there
is a dynamic change in arsenic speciation at the nZVI surface
on top of continuous arsenic sequestration from the aqueous
phase. Additional experiments are needed to elucidate the
individual reaction rates and the interrelations among these
processes.
Conclusion
Simultaneous oxidation and reduction of As(III) and the
reduction of As(V) to As(III) and As(0) on nZVI surfaces have
been observed. The elusive As(0) species suggested in previous
studies was clearly detected. The reactions can be reasonably
explained on the basis of the core-shell structure of nZVI, in
which the oxide outer layer and metallic core bestow distinctive
mechanisms for arsenic removal. The results shown here may
help in the understanding of the reaction mechanisms and the
stability of arsenic on ZVI surfaces and lead to better design of
arsenic sequestration technologies. In particular, in view of the
conventional adsorptive methods, which are critically affected
by arsenic speciation, concentration, and pH, the multifaceted
functionality of nZVI exhibited here may offer a more robust
and effective method for arsenic removal.
Acknowledgment. This work has been supported by grants
awarded to W.-x.Z. by the U.S. Environmental Protection
Agency (EPA STAR Grants R829625 and GR832225) and the
National Science Foundation (Grant No. 0521439), and to
B.E.K. by the National Science Foundation (Grant No. 0616644).
The authors also acknowledge support of this work by the
Pennsylvania Infrastructure Technology Alliance (PITA). The
authors gratefully acknowledge A. A. Herzing and C. J. Kiely
for providing the TEM images shown in Figure 2.
Supporting Information Available: The Supporting Infor-
mation contains a detailed description of the experimental
methods utilized in these studies of the reaction of As(III) and
As(V) with nZVI in solution. It provides experimental param-
eters and procedures used in the acquisition and curve fitting
of the XPS spectra. Table S1 gives the surface composition of
Letters
samples analyzed by XPS, and Table S2 provides solution pH
and Eh values before and after reactions. Figure S1 reports
solution pH values for various initial As(III) concentrations,
indicating the equilibrium conditions at which H3AsO3 dissoci-
ates into H2AsO3-. Figure S2 provides O1s HR-XPS spectra
obtained for different concentrations of As(III) and indicates
the different species of oxygen that form the oxide thin film
encapsulating the metallic iron core in nZVI. Figure S3 shows
As3d HR-XPS spectra from nZVI particles reacted with 100
mg/L As(III) for different reaction times. Figure S4 gives SEM
images of fresh and reacted nZVI samples. This material is
available free of charge via the Internet at http://pubs.acs.org.
References and Notes
(1) World Health Organization. Guidelines for drinking water quality,
Vol1: Recommendations, 2nd ed.; World Health Organization: Geneva,
Switzerland, 1993.
(2) U.S. Environmental Protection Agency, National Primary Drinking
Water Standards, 2003, EPA 816-F-03-016.
(3) Welch, A. H.; Westjohn, D. B.; Helsel, D. R.; Wanty, R. B. Ground
Water 2000, 38, 589.
(4) Smedley, P. L.; Kinniburgh, D. G. Appl. Geochem. 2002, 17, 517.
(5) Manning, B. A.; Hunt, M. L.; Amrhein, C.; Yarmoff, J. A. EnViron.
Sci. Technol. 2002, 36, 5455.
(6) Kanel, S. R.; Greneche, J. M.; Choi, H. EnViron. Sci. Technol. 2006,
40, 2045.
(7) Scott, M. J.; Morgan, J. J. EnViron. Sci. Technol. 1995, 29, 1898.
(8) Hug, S. J.; Leupin, O. EnViron. Sci. Technol. 2003, 37, 2734.
(9) Goldberg, S.; Johnston, C. T. J. Colloid Interface Sci. 2001, 234,
204.
(10) Mohan, D.; Pitman, C. U. J. Hazard. Mater. 2007, 142, 1.
(11) Su, C. M.; Puls, R. W. EnViron. Sci. Technol. 2001, 35, 1487.
(12) Kanel, S. R.; Manning, B.; Charlet, L.; Choi, H. EnViron. Sci.
Technol. 2005, 39, 1291.
(13) Leupin, O. X.; Hug, S. J. Water Res. 2005, 39, 1729.
(14) Sun, Y. P.; Li, X. Q.; Cao, J. S.; Zhang, W. X.; Wang, H. P. AdV.
Colloid Interface Sci. 2006, 120, 47.
(15) Li, X. Q.; Zhang, W. X. Langmuir 2006, 22, 4638.
(16) Moulder, J. F.; Stickle, W. F.; Sobol, P. E.; Bomben, K. D.
Handbook of X-ray Photoelectron Spectroscopy; Perkin-Elmer Corporation:
Eden Prairie, MN, 1992.
(17) Lackovic, J. A.; Nikolaidis, N. P.; Dobbs, G. M. EnViron. Eng.
Sci. 2000, 17, 29.
(18) Martin, J. E.; Herzing, A. A.; Yan, W. L.; Li, X. Q.; Koel, B. E.;
Kiely, C. J.; Zhang, W. X. Langmuir 2008, 24, 4329.
(19) Nurmi, J. T.; Tratnyek, P. G.; Sarathy, V.; Baer, D. R.; Amonette,
J. E.; Pecher, K.; Wang, C. M.; Linehan, J. C.; Matson, D. W.; Penn, R. L.;
Driessen, M. D. EnViron. Sci. Technol. 2005, 39, 1221.
(20) Manning, B. A.; Goldberg, S. EnViron. Sci. Technol. 1997, 31, 2005.
(21) Joo, S. H.; Feitz, A. J.; Waite, T. D. EnViron. Sci. Technol. 2004,
38, 2242.
(22) Katsoyiannis, I. A.; Ruettimann, T.; Hug, S. J. EnViron. Sci.
Technol. 2008, 42, 7424.
(23) Manning, B. A.; Fendorf, S. E.; Goldberg, S. EnViron. Sci. Technol.
1998, 32, 2383.
JP9051837